Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
Secure and optimize development budgets on
projects
Olivier is IAC's reference on reducing development
costs for projects. He leads missions in many
industrial sectors such as aerospace, defense,
medical, rail transport, etc.
Olivier Saint-Esprit
[email protected]+33 (0)6 28 72 07 67
Customer references:
2
Focus on development costs in aeronautics and defense
SPACE DEFENSE AERONAUTICS CONSUMER GOODS
Telecom Payload Radar System Power Distribution System Household appliance
Non-recurring costs 30 M€ 100 M€ 20 M€ 5 M€
Recurring unit cost 1 M€ 0,5 M€ 50 k€ 50 €
Annual production volume 2 units 10 units 20 units 500 000 units
Total production time 5 years 20 years 30 years 5 years
Cumulative recurring costs 10 M€ 100 M€ 30 M€ 125 M€
Thus, in aeronautics, space and defense, development budgets usually represent an average of 40%
to 80% of the total cost that a company must incur to develop and then manufacture a new system.
As a comparison, the ratio of non-
recurring costs sharply decreases for
less capital-intensive industries such as
consumer goods.
In industrial sectors with low production volumes, such as space, defense and aeronautics, medical
imaging or high-tech equipment, the amount of non-recurring costs1 spent by an industrial on a
new program is frequently equivalent to or greater than the sum of recurrent manufacturing costs
accumulated over the entire production life of the equipment.
The table below provides some representative orders of magnitude by sector of activity.
75%Space
Defense
Aeronautics
Consumer Goods
25%
50%50%
40%60%
4%96%
Non-recurring costs Cumulated recurring costs
1 Non-recurring costs refer to expenses incurred during the development of a project, to design, test and prototype a new system or equipment. These costs are spent only once, unlike the recurrent manufacturing costs that need to be incurred for each unit during the serial production phase.
3
Initial budget Budget at completion
GALILEO Satellite positioning system € 5bn € 13bn6
A400M ATLAS Military transport aircraft € 20bn€ € 31bn7
B787 Long-haul aircraft $ 5,8bn $ 32bn88
F-35 Lightning II Multirole combat aircraft $ 233bn > $ 400bn99
The typical cost structure of a product
development in aeronautics and defense is
presented opposite.
Engineering costs usually represent 70% of total
non-recurring costs2.
70%Engineering
Industrialisation
Risk provision
Prototypes
Tests
13%
10%
4%
3%
Securing these budgets is a major challenge
for industrial groups, as any drift affects the
company’s profitability and cash flow. In some
extreme cases such as in the nuclear power
sector, historical players have found themselves
in bankruptcy due to technological failures and
uncontrolled delays3.
A company's competitiveness regarding its
development expenditure is therefore crucial.
It is a matter of winning contracts and preparing
for the future, by preserving the company’s
capacity to carry out self-financed research
activities.
Nevertheless, most of the flagship programs in
aeronautic, space and defense are experiencing
budget and time overruns4.
For several reasons, notably related to increasing
technological complexity of equipment5
many programs are indeed seeing their initial
development budget explode:
2 Certification costs are not considered here; still they represent an important cost in aviation.
3 https://www.lecho.be/entreprises/energie/La-debacle-des-geants-du-nucleaire-Westinghouse-et-Areva/9880880
4 http://www.cad-magazine.com/sites/default/files/articles/pdf/cad192_pp16-18_repere%20siemens.pdf
5 Sophie LEFEEZ, « Toujours plus chers ? Complexité des armements et inflation des coûts militaires », IFRI Focus stratégique, n° 42, février
2013 https://www.ifri.org/sites/default/files/atoms/files/fs42lefeez.pdf
6https://www.latribune.fr/entreprises-finance/industrie/aeronautique-defense/combien-va-couter-le-programme-galileo-a-leurope-546146.html
7 https://fr.wikipedia.org/wiki/Airbus_A400M_Atlas#Retards_et_surco%C3%BBts
8 https://www.seattletimes.com/business/boeing-celebrates-787-delivery-as-programs-costs-top-32-billion/
9 https://www.usinenouvelle.com/article/le-couteux-chasseur-f-35-de-lockheed-martin.N498089
4
01 04
02 05
04
03
In the absence of well-established time reporting methods, information about the full cost of a project and the detail of these costs is not always available nor well documented. This difficulty is reinforced when significant volumes of hours have been outsourced externally;
Despite this situation, companies often devote less attention to securing their development costs, compared to efforts dedicated to optimizing the recurring costs of their products.
Several factors explain this state of affairs:
Few feedback and root cause analysis exercises are actually conducted. When they are, most of the projects concerned are those that have gone off track. Good practices implemented on projects that have met their commitments are rarely clearly highlighted, as compliance with initial estimates is considered as normal;
Finally, it should be pointed out (let’s point out?) that the optimization of study budgets remains a taboo in some companies because engineering teams are still reluctant to apply to themselves the principles of efficiency inherited from Lean Manufacturing, now well integrated by production teams12.
More and more sophisticated predictive methods are now available to evaluate the manufacturing costs of a product (see our recent article on this subject: Predictive costing for your competitiveness10), but these methods are still deficient to estimate a project’s study volumes;
Study costs are sometimes knowingly underestimated at the beginning of a project to win the investment decision with acceptable profitability11. When these biases are noted, the lack of reference points caused by the deficiencies listed in the previous points limits the possibilities of objection;
10 https://www.interactionconsultants.com/en/publication/predictive-costing-to-serve-your-competitiveness
11 The case of the A400M is emblematic of a voluntary underestimation accepted by all parties. When you look at the functions requested on the one hand, and the time and cost that the parties have committed to on the other, "the product is impossible," says one specialist. Valérie Lion, " A400M rattrapé au vol ", L'Express, n°25, 24-30 June 2011, p. 84. 86
12 Lean Engineering is a variation, intended for engineering activities, of the technical principles derived from Lean Manufacturing: to produce exactly the necessary knowledge at the right time (knowledge being perishable), without jolts or unnecessary "stocks" and without wasting time in activities without added value.
5
Nevertheless, an increasing number of companies choose to dedicate more attention to accurate estimation and securitization of their development budgets.
The first 2 steps are then essential to any progress initiative. They consist in ensuring the correct dimensioning of the initial provision, and especially in securing this initial provision to avoid slippages during the execution phase. This comes well before considering the reduction of the amounts engaged.
The estimation of the most accurate budget upstream of a project is the first key to competitiveness. Indeed, a budget that is poorly estimated or insincere bears the seeds of future overruns.
Frank Freiman has modelled the impacts of a poor budget estimation for a project, both upward and downward. The curve that bears his name illustrates that:
• when costs are underestimated, initial manufacturing and planning plans prove impractical. The actual cost then explodes as the project must be reoriented. It results in unplanned additional work, reorganizations and need of additional resources.
• the overestimation of costs is self-fulfilling, leading to waste
1 = Underestimates lead to disaster2 = Realistic estimates minimize final costs
3 = Overestimates become self-fulfilling prophecies
Figure by MIT OCW. Adapted from Freiman, F. R. “The Fast Cost Estimating Models.” AACE
Transactions (1983).
Estimated Cost
0
1 2 3
0
10
10
20
20
The Freeman Curve
Fina
l Pro
grm
Cos
t
66
Despite the quality of the initial estimate,
the cost drift factors of a project remain
important. The missions conducted in recent
years by the IAC teams have allowed to
identify 3 major causes to cost drifts of a
project after its launch. It is a matter of:
Changes in the expression of needs after the start of the project;
Frequently insufficient evaluation of maturity on the technological bricks implemented on the project;
Incorrect estimation of the actual carry over level (or reuse) of solutions from one program to another.
Let’s develop these 3 points in the rest of this
article.
01
02
03
77
1. Evolutions in the expression of needs during a project are the source of many slippagesIn complex system development projects, such
as railway or aeronautical programs for example,
changes frequently occur. Of course, it would
be illusory to imagine abolishing changes
completely because they are intrinsic to
concurrent engineering project methodologies.
They allow for instance to begin the study of
the fuel system on an aircraft program without
waiting for the certification of the engine.
Flexibility is also needed to adapt to late market
developments, for example in the space sector
where development cycles are increasingly
shorter.
Moreover, design work is sometimes initiated
“ahead of schedule” in anticipation of
formalization of the customer need – in certain
cases even before the official notification
or launch of the project. This anticipation is
made in the hope of “buying time” or freeing
up calendar margins on a schedule deemed too
constrained. Although based on good intentions,
these anticipations frequently result in a
particularly high scrap14 or rework15 rate. They
can even lead to impossibilities such that the
project cannot succeed, as it was the case for
the LOUVOIS payroll software launched in 1996
and definitively abandoned in 2013 without ever
having functioned properly16.
However, unpredictable specification changes
remain marginal: the review of specifications in
the preliminary design phase makes it possible
to know their sensitivity and quantify their
evolution risk.
The best approach then consists in anticipating
the occurrence of modifications: adjusting
the development logic accordingly and
communicating to the customer and internal
entities potentially generating modification
requests, the dates by which they must have
been notified.
The impact of a specification evolution during
development depends on 3 parameters:
a) progress of the program
b) typology of the business activities affected
(system engineering, mechanical design,
hardware or software electronic design, etc.)
c) interdependencies between departments.
13 Golub's Law No. 2, a famous unknown computer scientist.
14 Scrap: waste: activities carried out are useless for the project, and the time spent is a dry loss.
15 Rework: by integrating additional activities not initially planned (thus generating additional costs), the work can be reoriented to meet the needs of the project.
16 "It also appears that the functional design of this information system was insufficient to model such complexity. Indeed, the general functional specifications were not even written with enough acuity even though the realization of the software had started. In 2015, four years after the first deployment, the delegated project owner had to write functional specifications of allowances in a hurry, even though they have been calculated since deployment. Unique joint pay software. (2018, February 18). In Wikipedia. Retrieved from https://fr.wikipedia.org/w/index.php?title=Logiciel_
"One of the advantages of setting vague objectives for a project is that you will have no difficulty estimating the corresponding expenses...13".
88
The flexibility matrix presented below is an educational tool that can be used to explain the level of
impact of a specification evolution as the project milestones are crossed.
This matrix can be implemented not only to improve communication between internal teams, but also
proactively for the external client, to highlight the impact of a change in the expression of need and
to contract in advance the additional cost that will be invoiced in case of late evolution.
This tool can also be deployed for suppliers, to make the entire development process more flexible
between project partners.
Low impact Medium impact Strong impactNo impact
Impact on budget and planning
MECHANICAL design
HARDWARE design of
electronic boardsWIRING
SOFTWARE modification of
function
Addition of a SOFTWARE
function
Before KOM
No Impact if relocation of mechanical
interfaces (same volume & shape)
No impact if up to 10 signals
modification on ICD
No impact if:- Pin allocation
modification- Gauge
modification- Routing
modification- Add/removal of wire in the
perimeter of the defined connector
No impact if modification can be achieved in
current processing ressources
AFTER KOM
AFTER PDR B1 standard
AFTER CDR B1 standard
AFTER PDR B2 standardAFTER B1 standard deliveryAFTER CDR B2 standardAFTER B2 standard delivery AFTER B2 standard Flight Clearance
AFTER CDR C-model
After C-model delivery
Enginering department impacted by a change in specification
Prog
ram
mile
ston
es
Example of a flexibility matrix for a program to develop an aeronautical
electrical generation system
9
Average Program Research, Development, Test, and Evaluation Cost Growth from First Full Estimate
2. Assessing the maturity of the technological bricks implemented on the project is also crucialIndeed, a development program is generally built
around Key Technical Elements17 made available by
the company's upstream research departments.
Controlling the level of technological maturity
of these bricks is one of the success factors
of the project.If the development planning is
based on the use of mature bricks, and they
are not in fact, then they prove unusable by the
development teams and additional work must
urgently be carried out.
One of the major causes of budget drift lies
in insufficient prepared incorporation of these
bricks into the program schedules.
The United States Government Accountability
Office (GAO) analyzed 52 weapons programs. It
revealed that most of these programs started
with lower levels of knowledge and maturity than
recommended by best practice. The table18 below
shows that programs that started with immature
bricks recorded an average 34.9% increase in
their development budgets, compared to only
4.8% for projects supported by mature bricks.
Mature technologies
4,8
%
34,9
Immature technologies
0
10
20
30
40
17 A technical element is said to be "key" if the product or system considered for evaluation depends on this technical element to achieve its main functionalities with an acceptable cost and development time, as well as production costs, and if this technical element or its implementation is intrinsically new or only new in their implementation environment.
18 Office, U. S. G. A. (2006). Defense Acquisitions: Assessments of Selected Major Weapon Programs, (GAO-06-391). Retrieved from https://www.gao.gov/assets/250/249468.pdf
“What gets us into trouble is not what we don't know. It's what we know for sure that just ain't so…” Mark Twain
10
Knowledge of the TRL scale (Technology Readiness
Level-developed by NASA in 1974 to provide a
common basis for defi ning the level of maturity of
a technology) is now widespread, but we note that
its eff ective use during the planning and budgeting
phases of projects remains insuffi cient.
Inappropriate implementation of this indicator can
also lead to heavy additional costs to projects
budgets because despite the linearity of this scale,
the eff orts to exceed level 6 are very substantial19.
Real system completed and qualifi ed by successful operational missions
Real system completed and qualifi ed by tests and demonstrations
Demonstration of a prototype system in an operational environment
Demonstration of a prototype or model system/subsystem in a representative environment
Validation of components and/or models in a representative environment
Validation of components and/or models in the laboratory
Analytical or experimental evidence of the main functions and/or characteristics of the concept
Technological concept and/or formulated applications
Basic principles observed or described
TRL 9
TRL 8
TRL 7
TRL 6
TRL 5
TRL 4
TRL 3
TRL 2
TRL 1
System test, launch and
reindustrialisation
System/subsystem
development
Technology Development
Basic technological
research
Technology Demonstration
Research and demonstration of
feasibility
TRL scale from DGA's 2009 Defense and Security Research and Technology
Strategic Plan 20
19 "Actually, there is a huge distance between being technically proven and implementing successfully. This diffi culty in the transition of a new technology is aff ectionately called "Death Valley." Boeing: Technology Readiness and the Valley of Death. (n.d.). Accessed April 11, 2018, at http://www.boeing.com/features/innovation-quarterly/may2017/feature-thoughtleadership-newman.page
20 Armaments Portal: The Strategic Research and Technology Plan (SP R&T) 2009. (n.d.). Retrieved 11 April 2018, from https:// www.ixarm.com/sites/default/fi les/documents/PS_R_T_english_web_0.pdf
11
We have identified 3 main reasons why errors are made in assessing the maturity of key technical
elements despite the use of the TRL scale:
01
03
02Misunderstandings persist frequently between teams on the actual level of
maturity as the R&T teams in charge of the
study and the provision of the technological
bricks are not the ones who carry out the
development: an organizational solution to
this pitfall then resides in the constitution of
integrated Research ↔ Development teams;
Finally, when TRL and/or MRL levels are correctly assessed at the beginning of the project, the development logic and the detailed planning of the resulting activities are not always constructed to
ensure a gradual rise in maturity. A flowchart that highlights the contribution of each activity to this rise
in maturity (see diagram below) allows to organize development accordingly, for example by providing
upstream risk mitigation activities, or by "taking" an activity off the critical path.)
The TRL rating alone is not sufficient to account for the incorporability of a brick within a development: other
indicators must be assessed, at least the MRL21,
or even other indicators capable of restoring
the ease of integration of a component within
a complex system22;
21 There is an established Manufacturing Readiness Level (MRL) scale, adopted by the U.S. Department of Defense in 2005, that addresses the feasibility and affordability of producing the technology at the required scale and rate. There are also integration maturity metrics used to assess an Integration Readiness Level (IRL) scale. And both contribute to a System Readiness Level (SRL). Boeing: Technology Readiness and the Valley of Death. (s. d.). Accesed 11th April 2018, on http://www.boeing.com/features/innovationquarterly/may2017/feature-thought-leadership-newman.page
22 See on this subject: Garg, T., Eppinger, S., Joglekar, N., & Olechowski, A. (2017). Using TRLs and system architecture to estimate technology integration risk. Accessed online 11 April 2018 at: http://web.mit.edu/eppinger/www/pdf/Garg_ICED2017.pdf
ActivitéB
ActivitéC
ActivitéD
ActivitéA
TRL 6
MRL 6 MRL 6
MRL 7 MRL 7
MRL 7
MRL 5
MRL 5
MRL 5
TRL 7
TRL 7 TRL 8
TRL 7 TRL 7
TRL 7TRL 6
ActivitéF
ActivitéE
A simple graphical representation of this type highlights the critical path(s) (in blue) of maturity rise on the different indicators selected - it should be pointed out that the level of maturity resulting from several activities is always equal to the lowest level of maturity.
12
3. The inaccurate estimate of the carry over level generates risks
Finally, it is the incorrect estimation of the actual
carry over (or reuse) level of the solutions from
one program to another that generates risks of
discrepancies between the initial forecast and
the actual figure.
Indeed, it is not often that a new development
is done entirely from scratch, without providing
for the re-use of a physical element or software.
Reuse or carry over assumptions are then
considered when budgeting activities, in order to
materialize the economy of the redevelopment
of these functions or subsets previously
developed. The most mature industries have
defined within the framework of Product Policy
reflections elementary bricks, intended to allow
the design with less delays and less costs for
future products in a logic of modular design
(we also speak of Platforming23 or sometimes
of versioning). In the energy and rail sectors,
the most advanced players have begun their
transition from the "Engineering-to-order24"
world to the "Configure-to-order25".
However, the carry-over of a hardware or
software solution does not spare the company
from re-expending a substantial budget to adapt
and incorporate the re-used solutions, if this
carryover was not thought through during the
initial design steps. Indeed, "technical difficulties
precisely lie in systems integration: the cost
and organizational difficulties are exponential
compared to the additional functions26",
which means that integrating a reuse solution
into a complex system can prove extremely
complex, and ultimately can cost as much as
a new development if this integration was not
initially planned. The Reuse Readiness Level
(RRL27) indicator proposed by NASA aims at
explaining the level of re-usability of a software
in a different application than the one it was
originally developed for.
23 https://www.iac.fr/publication/publication-platforming
24 Engineer-to-order (ETO) approach is one in which a company designs and manufactures a product based on very specific customer requirements. Engineered-to-Order (ETO) | Arena Solutions. (s. d.). Accessed 11th April 2018, on https://www.arenasolutions.com/ resources/articles/engineered-to-order/
25 Configure-to-order (CTO) represents the ability for a user to define the component make-up (configuration) of a product at the very moment of ordering that product, and a vendor to subsequently build that configuration dynamically upon receipt of the order. Configure To Order Manufacturing Software. (s. d.). Accessed 11th April 2018, on http://www.software4manufacturers.com/manufacturing/styles_configure_to_order.aspx?Styles_Shortname=CTO
26 Sophie LEFEEZ, "Always more expensive? Complexity of armaments and inflation of military costs", IFRI Strategic Focus, No. 42, February 2013 https://www.ifri.org/sites/default/files/atoms/files/fs42lefeez.pdf
27 Reuse Readiness Levels (RRLs) - Earth Science Data System Working Groups - ESDSWG - Earthdata Wiki. (n.d.). Accessed 11 April 2018, at https://wiki.earthdata.nasa.gov/download/attachments/49446977/RRLs_v1.0.pdf?version=1&modificationDate=1428606889506&api=v2
13
It is also the foundation of the Design for
Variety28 methodology, aimed at developing
standardized and modular product
architectures. This approach implements two
parameters:
• the Generational Variety Index (GVI) which
represents the necessary redesign intensity
to meet future market needs;
• the Coupling Index (CI) indicating the intensity
of the coupling between the components of
a product.
This methodology will be detailed in a
forthcoming IAC publication on modular
design methodologies.
Thus, we note that a sizable proportion
of budget overruns are related to a poor
upstream assessment of the actual re-
usability of solutions previously developed.
Careful attention to these aspects upstream
of a project together with the use of objective
evaluation tools enables to correctly assess
the reuse assumptions on which the project
budget is built.
28 Martin, M. V., & Ishii, K. (2002). Design for variety: developing standardized and modularized product platform architectures. Research in Engineering Design, 13(4), 213-235. Accessed 11th April 2018, on http://web.mit. edu/deweck/Public/ESD39/Readings/GVImetrics.MartinIshii-2002.pdf
14
To Conclude
The aeronautics, space, telecommunication
and defense sectors are now seeing their
development cycles shorten, due to the
emergence of new players and profound
upheavals in business models.
This acceleration combined with a technological
frenzy29 jeopardizes the ability of industrial
companies to realistically assess and secure
their development budgets.
If numerous factors of uncertainty can never be
eliminated such as the evolution of need during
the project or the development difficulties of
a new technology, good practices consist in
assessing the risk levels incurred as soon as
possible and with an objective approach. When
carried out upstream, these preparatory works
turn out to be a very good investment, helping
to avoid heavy slippages. They moreover open
the door to a serene progress of the project by
increasing the level of confidence of the team
in the submitted budget.
29 "The F-35 Joint Strike Fighter contains an "crazy number" of chips according to a semiconductor expert." The Hunt for the Kill Switch", IEEE spectrum magazine, May 2008. quoted in Sophie LEFEEZ, "Ever more expensive? Complexity of armaments and IFRI Strategic Focus, No. 42, February 2013 https://www.ifri.org/sites/default/files/atoms/files/fs42lefeez.pdf
Methods to optimize project execution such as Lean Engineering methodologies can then be deployed. They will be the subject of a future publication.
15